O R I G I NA L A RT I C L E
doi:10.1111/j.1558-5646.2011.01557.x
RATES OF MORPHOLOGICAL EVOLUTION ARE
CORRELATED WITH SPECIES RICHNESS IN
SALAMANDERS
Daniel L. Rabosky1,2,3 and Dean C. Adams4,5
1
Department of Integrative Biology, University of California, Berkeley, California 94750
2
Museum of Vertebrate Zoology, University of California, Berkeley, California 94720
3
E-mail: drabosky@berkeley.edu
4
Department of Ecology, Evolution, and Organismal Biology, Iowa State University, Ames, Iowa 50011
5
Department of Statistics, Iowa State University, Ames, Iowa 50011
Received July 26, 2011
Accepted November 24, 2011
Data Archived: Dryad doi:10.5061/dryad.vt41c78j
The tempo and mode of species diversification and phenotypic evolution vary widely across the tree of life, yet the relationship
between these processes is poorly known. Previous tests of the relationship between rates of phenotypic evolution and rates
of species diversification have assumed that species richness increases continuously through time. If this assumption is violated,
simple phylogenetic estimates of net diversification rate may bear no relationship to processes that influence the distribution
of species richness among clades. Here, we demonstrate that the variation in species richness among plethodontid salamander
clades is unlikely to have resulted from simple time-dependent processes, leading to fundamentally different conclusions about
the relationship between rates of phenotypic evolution and species diversification. Morphological evolutionary rates of both size
and shape evolution are correlated with clade species richness, but are uncorrelated with simple estimators of net diversification
that assume constancy of rates through time. This coupling between species diversification and phenotypic evolution is consistent
with the hypothesis that clades with high rates of morphological trait evolution may diversify more than clades with low rates. Our
results indicate that assumptions about underlying processes of diversity regulation have important consequences for interpreting
macroevolutionary patterns.
KEY WORDS:
Adaptive radiation, macroevolution, morphological evolution, phylogenetics, speciation.
Explaining the variation in both species richness and phenotypic
diversity across the tree of life is a major challenge in evolutionary biology. Previous studies have demonstrated that rates of
both species diversification and phenotypic evolution vary widely
among clades (e.g., Alfaro et al. 2009; Eastman et al. 2011;
Venditti et al. 2011), but we are only beginning to directly investigate the potential relationships between these patterns and the
processes that generate them (Adams et al. 2009). There are many
reasons to expect positive correlations between rates of phenotypic evolution and rates of species diversification. For example,
high rates of phenotypic evolution might lead to high speciation
C
1807
rates because the capacity for rapid local adaptation might lead
to rapid evolution of prezygotic isolation mechanisms between
populations. Indeed, the theory of “punctuated equilibrium” was
proposed in part to explain the observation that phenotypic change
in the fossil record frequently appears to be associated with speciation (Eldredge and Gould 1972).
The capacity for rapid phenotypic evolution may directly
facilitate species diversification by increasing the ability of a radiating clade to exploit ecological opportunity (Parent and Crespi
2009; Slater et al. 2010; Martin and Wainwright 2011). Clades
with high rates of phenotypic evolution should have increased
C 2012 The Society for the Study of Evolution.
2012 The Author(s). Evolution Evolution 66-6: 1807–1818
DA N I E L L . R A B O S K Y A N D D E A N C . A DA M S
ability to explore ecological space and should diversify more than
clades with low phenotypic rates if there are strong ecological
controls on species richness. The idea that such “evolvability”
might promote species diversification via ecological mechanisms
has a long history in evolutionary biology and is fundamental
to understanding the ecological mechanisms underlying key evolutionary innovations (Liem and Osse 1975; Heard and Hauser
1995). Vermeij (1973a,b) proposed that groups with high evolutionary lability in form (“versatility” sensu Vermeij 1973a) have
replaced less-labile groups through time, perhaps due to the ability of high-lability groups to better utilize a broader spectrum
of available resources (e.g., by increasing the overall size of the
realized adaptive zone: [Vermeij 1973a]).
Although correlations between rates of phenotypic evolution
and rates of species diversification may be expected a priori, few
studies have directly investigated this question. Several studies
have demonstrated that rates of phenotypic evolution may be elevated during periods of time when rates of species diversification
are highest (Harmon et al. 2003; Mahler et al. 2010), but these
results do not necessarily imply that lineages with high rates of
phenotypic evolution diversify more than lineages with low rates
of phenotypic evolution. Ricklefs (2004) suggested that morphological disparity and species diversity were correlated across major clades of passerine birds, but subsequent analyses indicated
that such correlations can potentially be attributable to variation
in the ages of clades alone (e.g., with no variation in either rates
of species diversification or phenotypic evolution; Purvis 2004;
Ricklefs 2006).
Recently, Adams et al. (2009) used a phylogenetically explicit approach to estimate rates of phenotypic evolution in clades
of plethodontid salamanders. They tested whether these phenotypic rates were correlated with clade-specific estimates of net
diversification rates, which were computed from the estimated
crown clade ages and extant species richness (Magallon and
Sanderson 2001). Despite considerable among-clade variation in
rates of both species diversification and phenotypic evolution,
they found no evidence for a general relationship between phenotypic evolutionary rates and rates of species diversification. Rates
of body size and shape evolution were uncorrelated with net rates
of species diversification across the 15 clades they considered.
The analysis in Adams et al. (2009) assumed that net rates of
species diversification could be inferred for each salamander clade
from the ages of those clades in conjunction with their standing
species richness. The estimators used in this and many other
studies make the simple assumption that clades begin radiating
from one or two ancestral lineages (for stem and crown clades,
respectively), and that speciation and extinction rates have been
constant in time throughout the entire history of the clade. These
estimators are termed constant-rate (CR) estimators. Thus, for
a clade of N extant species beginning with a single ancestral
1808
EVOLUTION JUNE 2012
lineage some t years before the present and in the absence of
extinction, the net diversification rate is simply log(N)/t. Almost
all published analyses of diversification in higher taxa have used
variations on this basic equation, with the specific choice of CR
estimators differing only in assumptions about extinction rates,
the number of ancestral lineages, and whether the estimates of
net diversification rates are conditioned on clade survival to the
present (Magallon and Sanderson 2001; Wiens 2007; Alfaro et al.
2009). The theory underlying the CR estimators (Bailey 1964;
Raup 1985) can easily be extended to scenarios where rates vary
through time (Rabosky 2009b,2010b), although this is rarely done.
By applying CR estimators to species richness within clades,
researchers explicitly assume that diversity is a function of the
net rate of species accumulation through time. An alternative
approach is not to assume rate constancy, but rather to model
species diversification under some nonconstant process of diversification through time. One candidate process involves diversity
dependence of speciation and/or extinction rates, an idea with a
rich history in the paleontological literature (Rosenzweig 1975;
Sepkoski 1978; Walker and Valentine 1984; Alroy 2008). In one
of the earliest paleobiological applications of computer simulation, Raup et al. (1973) compared patterns of clade diversification
in the fossil record to those generated under a null model that
explicitly included diversity equilibria. Their use of an equilibrium diversity model was motivated in part by a general sense
that diversity dependence provided a far better description of empirical patterns than a model of unconstrained diversity increase
through time. More recently, a number of neontological studies
have suggested that temporal patterns of speciation in molecular
phylogenies are consistent with diversity-dependent regulation
of speciation–extinction dynamics (Weir 2006; McPeek 2008;
Rabosky and Glor 2010; Etienne et al. 2011).
Several studies have demonstrated that distributions of
species richness across clades are inconsistent with a CR diversification process and suggestive of diversity-dependent control
(Ricklefs 2007; Rabosky 2010b; Vamosi and Vamosi 2010) or
extinction-driven clade dynamics (Pyron and Burbrink 2011).
Because diversity dependence can lead to a breakdown of the
relationship between clade age and species richness, CR estimators may simply covary with clade age in a manner that is
largely decoupled from the factors that determine clade diversity
(Rabosky 2009a). To better understand the factors that influence
species richness within clades, it is important to test whether CR
estimators provide a reasonable estimate of clade diversification
histories before using CR-based estimates in subsequent analyses.
Here, we examine the relationship between clade diversification and phenotypic evolutionary rates in plethodontid salamanders. The premise of our article is that the validity of CR
estimators cannot be assumed a priori, but must be tested by
researchers before examining the relationship between species
S P E C I E S D I V E R S I F I C AT I O N A N D P H E N OT Y P I C E VO L U T I O N
diversification and other factors. If species richness within clades
cannot be modeled adequately as a time-dependent process, then
it is not appropriate to analyze and/or compare CR diversification
estimates for clades. A major goal of diversification studies is
to understand the factors that influence species richness within
clades, and the relevant question is thus whether we should “correct for time” by computing net diversification rate estimates for
clades or whether we should directly investigate correlations between clade richness and phenotypic/ecological covariates.
In addition, we also investigate a recent test proposed to assess the validity of CR estimators (Wiens 2011) and find that it is
characterized by unacceptable error rates. We then apply a hierarchical Bayesian model for describing patterns of diversification
rate variation among clades, and we use posterior predictive simulation to test whether this model can explain the relationship
between age and diversity in the salamander dataset. We modeled the dynamics of speciation through time within salamander
clades to test whether temporal declines in diversification rates
could explain the observed decoupling between age and species
richness. We explicitly compare the predictive ability of CR rate
estimates and log-transformed richness in explaining the variation
in phenotypic evolutionary rates among salamander clades. Our
results suggest that CR estimators applied to plethodontid salamanders are not valid and that log-transformed species richness
is a more appropriate summary of clade diversification. We find
strong support for coupling between species richness and rates of
phenotypic evolution.
Materials and Methods
DATA AND ESTIMATION OF PHENOTYPIC RATES
Phylogenetic data and morphological rate estimates were taken
from Adams et al. (2009). The data include a time-calibrated
molecular phylogenetic tree for 191 species of plethodontid salamanders comprising 15 focal clades, and estimates of species richness and crown ages for each clade. The plethodontid phylogeny
was estimated from maximum likelihood analysis of nuclear and
mitochondrial DNA and made ultrametric using penalized likelihood; full details are available in Adams et al. (2009) and Kozak
et al. (2009).
Rates of both body size and shape evolution were obtained
from morphometric analysis of 190 species distributed across the
focal clades (Adams et al. 2009). Briefly, seven standard morphometric variables were measured for each of 1573 adult salamander
individuals (a mean of 6.7 individuals per species). A principal
components analysis (PCA) was performed on the covariance matrix of log-transformed measurements for individuals, and mean
PC scores were computed for each species. Because all variables
had comparable and positive loadings on PC1, this variable was
treated as an overall index of body size. PC2–PC7 were treated
as indices of body shape. Species’ mean scores for each principal
component were used to estimate rates of phenotypic evolution
under a Brownian motion process. The estimated evolutionary
rate for PC1 was taken as an index of the rate of size evolution.
A matrix of shape rates, along with their estimated covariances,
was estimated from PC2–PC7 scores. The sum of the diagonal elements of this matrix was taken as an overall estimate of the rate
of body shape evolution (McPeek et al. 2008). These estimates
of evolutionary rates are independent of the number of species
sampled from each clade as well as the amount of time available for diversification in each clade, unless rates themselves are
correlated with clade richness and/or if a model of time-invariant
Brownian motion fails to describe the data.
Three lines of evidence suggest that our phenotypic evolutionary rates are valid. First, a simple Brownian motion process
provides a reasonable fit to both size and shape rates of phenotypic
evolution and outperforms an Ornstein–Uhlenbeck (OU) model
for 13 of 15 clades (Adams et al. 2009). Second, if phenotypic rates
are confounded with clade age, perhaps due to constraints or the
scaling of rates with timescale of measurement (Gingerich 2001),
then we should observe negative correlations between phenotypic
rates and clade age. We tested for such a relationship using phylogenetic generalized least-squares regression (Martins and Hansen
1997), finding no significant relationship between either clade age
and size rate (Fig. 1A; β = 0.001; P = 0.36) or between clade age
and shape rate (Fig. 1B; β = 2.4 × 10−5 ; P = 0.46).
Finally, if our phenotypic evolutionary rates reflect the rate at
which clades accumulate morphological variance through time,
then phenotypic disparity should be positively correlated with
clade age. We estimated phenotypic disparity for each clade
as the sum of squared Euclidean distances between measured
specimens, standardized by the sample size. This measure of
phenotypic disparity was significantly and positively correlated
with clade age (Fig. 1C; phylogenetic generalized least-squares
[PGLS] β = 0.025; P = 0.05). Because mean pairwise phylogenetic (patristic) distance between species within clades is highly
correlated with clade age (Spearman’s r = 0.91; P < 0.001),
these results imply that older clades occupy a greater volume of
phenotypic space than young clades. The PGLS results reported
above (Fig. 1C) differ slightly from those reported in Adams et al.
(2009), because here we modeled tip variances (e.g., the diagonal of the phylogenetic variance–covariance matrix) in our PGLS
regressions by the total root-to-tip distance for each clade. Results reported in Adams et al. (2009) modeled tip variances as the
root-to-crown age distance for each clade.
TESTING THE VALIDITY OF CR DIVERSIFICATION
ESTIMATORS
For consistency throughout, we denote the per-lineage speciation rate by λ, the per-lineage extinction rate by μ, the net
EVOLUTION JUNE 2012
1809
DA N I E L L . R A B O S K Y A N D D E A N C . A DA M S
B
0.10
0.9
0.002
Shape rate
Size rate
0.08
C
0.06
0.04
Disparity
A
0.001
0.6
0.3
0.02
0.00
0.000
10
15
20
25
30
35
0.0
10
15
Clade age (Ma)
20
25
30
Clade age (Ma)
35
10
15
20
25
30
35
Clade age (Ma)
Relationships between clade age and (A) rates of body size evolution, (B) rates of body shape evolution, and (C) phenotypic
disparity. There is no relationship between clade age and either of the morphological evolutionary rates, but phenotypic disparity is
Figure 1.
significantly and positively associated with clade age.
diversification rate (λ – μ) by r, and the relative extinction rate
(μ/λ) by ε. Wiens (2011) proposed that the correlation between
clade species richness (N) and clade diversification rate (log(N)/t
or variants thereof) could provide a simple test for the validity of
CR diversification estimators. If this correlation is positive, then
diversification rates provide some predictive ability with respect
to clade richness, and CR estimators (Magallon and Sanderson
2001) retain explanatory power. However, there is a potential circularity with this test, as estimates of diversification rates are
computed directly from the species richness values. As a consequence of this mathematical relationship, the correlation analyses
may be compromised, as the association between species richness and clade diversification measures the relationship between
a single variable (N) and a composite measure of itself (log(N)/t)
(see Atchley et al. (1976) for a similar discussion).
To determine whether this was the case for this test, we performed the following simulation. First, we simulated 50-clade
datasets where we (1) sampled clade ages from a uniform distribution; (2) paired each of these ages with a species richness
value sampled from a geometric distribution; (3) used the sampled ages and richness values to compute the net diversification
rate, r, for each clade, using the method-of-moments estimator
(Magallon and Sanderson 2001) under several relative extinction rates (ε); and (4) we computed the rank-order correlation
between clade diversity and r for each simulated dataset. Thus,
clade ages and species richness values in the simulations are
sampled from uncorrelated distributions, and the species richness
of each clade is not a function of any time-dependent process
(by definition, the CR estimator describes a rate-limited process
and requires that species richness in clades is some function of
time). As expected, diversification rates and species richness are
strongly correlated, despite the fact that no biological relation-
1810
EVOLUTION JUNE 2012
ship between clade age and diversification rate was input into the
simulation process (Fig. 2). As clade richness in simulations is
sampled independently of clade age, a positive correlation implies
type I error. Here, net diversification rates bear no relationship to
clade diversity in any meaningful sense, yet based on the test of
Wiens (2011) we would conclude that diversification rates bear
some causal relationship to richness in a very high proportion of
simulations.
AGE-DIVERSITY RELATIONSHIPS IN SALAMANDERS
For the 15 clades analyzed by Adams et al. (2009) there is
no relationship between crown clade age and species richness
(Spearman’s ρ = –0.15, P = 0.59), suggesting that CR estimators
may provide a poor explanation for observed patterns of species
richness in clades and that diversity dependence may be constraining richness within clades (Ricklefs 2007; Rabosky 2010b). Although Rabosky (2009b) demonstrated that declining rates within
clades could eliminate the relationship between clade age and
species richness, it is also possible that this pattern could result from among-clade heterogeneity in diversification rates. This
scenario was explicitly considered in Rabosky (2010b), where
among-clade rate variation was modeled by assuming that clade
diversification rates were drawn independently from a lognormal
distribution. However, the Rabosky (2010b) model is explicitly
nonphylogenetic: it assumes that there is no phylogenetic signal in
the distribution of diversification rates across clades. In principle,
however, such phylogenetic autocorrelation in rates could further
weaken the expected relationship between clade age and species
richness, perhaps eliminating it altogether.
To address this, we utilized a phylogenetic counterpart to
the hierarchical model in Rabosky (2010b). We assume that
the logarithm of the diversification rate across clades follows a
S P E C I E S D I V E R S I F I C AT I O N A N D P H E N OT Y P I C E VO L U T I O N
ε=0
Type I Error
Frequency
A
-1.0
-0.5
0.0
0.5
1.0
B
f (θ, ν, r1 , . . . r N )
∞
∞
...
f (r1 , . . . r N |θ, ν) f (θ) f (ν) dr1 , . . . dr N ,
=
Frequency
ε = 0.95
0
-0.5
0.0
0.5
1.0
Correlation between diversity
and CR rate estimate
Correlations between net diversification rates and
species richness expected when species richness and clade age
are drawn independently from uncorrelated distributions devoid
Figure 2.
of biological significance. Net diversification rates were computed
under low (A) and high (B) relative extinction rates. The test proposed by Wiens (2011) to assess the validity of constant-rate (CR)
diversification estimators will almost always recover positive correlations when data lack any signal of a CR diversification process.
Because CR diversification estimators are computed from species
richness, they will generally be correlated with those richness values, even when diversity has not been generated under a timedependent diversification process.
multivariate normal distribution with a mean equal to the value
at the root node and a covariance structure specified by the phylogenetic variance–covariance matrix. This is simply a pairwise
matrix containing sums of shared branch lengths for all species
pairs. Diversification rates thus follow a multivariate lognormal
distribution. This is similar to the model used to describe phylogenetic autocorrelation of molecular evolutionary rates along the
branches of a phylogenetic tree (Thorne et al. 1998; Drummond
et al. 2006). There are two hyperparameters in the model: the
diversification rate at the root of the tree (θ), and a scalar multiplier of the phylogenetic variance–covariance matrix (ν). The
probability density of observing a clade with ni species, of age ti ,
is given by
0
0
∞
0
(2)
where f (θ) and f (ν) are prior densities on the root rate and scale
parameter, respectively. The joint posterior distribution is thus
-1.0
f (n i |θ, ν, ti )
∞
=
...
given its age, as a function of hyperparameters θ and ν, requires
integrating over all possible diversification rates at the tips of the
tree.
We adopted a Bayesian approach to perform the integration
in equation (1). This involves little more than specifying prior
distributions on the hyperparameters (θ, ν) and approximating the
joint posterior distribution of the hyperparameters using Markov
Chain Monte Carlo (MCMC). The joint prior distribution is given
by
f (n i |ri , ti ) f (ri |θ, ν, r1 , r2 , . . . r N ) dr1 , . . . dr N ,
(1)
where ri is the rate assigned to the ith terminal node. Computing
the probability of observing even a single clade richness value
f (θ, ν, r1 , . . . r N |n 1 . . . n N ) = f (θ, ν, r1 , . . . r N )
× f (n 1 , . . . n N |r1 , . . . r N )
(3)
and the data likelihood f (n1 . . . . nN | r1 . . . rN ) is computed as in
previous studies (Bailey 1964; Bokma 2003).
We assumed exponential prior distributions on both hyperparameters and used MCMC to summarize their joint posterior
distribution. We ran separate MCMC analyses of 106 generations
under six relative extinction rates (ε = 0, 0.2, 0.4, 0.6, 0.8, 0.99),
sampling parameters every 1000 generations. For comparison, we
also implemented a nonphylogenetic version of the hierarchical
model described above (see Rabosky (2010b) for a non-Bayesian
version). We simply assumed that net diversification rates for each
clade are drawn independently from a lognormal distribution with
(log) mean θ and variance ν. We refer to this latter model as the
uncorrelated lognormal model.
POSTERIOR PREDICTIVE SIMULATION
If the CR estimators provide a meaningful summary of clade diversification histories, they must be able to explain the lack of
relationship between clade age and species richness observed in
salamanders (Rabosky 2010b). We conducted posterior predictive
simulations for each combination of model and relative extinction rate to test whether phylogenetic and uncorrelated lognormal
models of diversification rate variation can explain the observed
distribution of species richness across salamander clades. We sampled hyperparameters θ and ν from their joint posterior distribution, then used these parameters to generate a sample of net diversification rates for each clade. For the phylogenetic model, this
entailed drawing rates from a multivariate lognormal distribution
with a covariance structure proportional to the observed phylogenetic variance–covariance matrix. We then simulated species
richness values for each clade, given the observed clade age, the
relative extinction rate, and the simulated diversification rate. For
EVOLUTION JUNE 2012
1811
DA N I E L L . R A B O S K Y A N D D E A N C . A DA M S
each simulated dataset, we computed the rank-order correlation
between clade age and log-transformed species richness.
SPECIATION RATES WITHIN CLADES
One possible explanation for a lack of relationship between clade
age and species richness is that net diversification rates have
slowed through time within clades, perhaps due to diversity dependence of speciation and/or extinction rates (Ricklefs 2007;
Rabosky 2009a, 2010b; McInnes et al. 2011). The ecological
mechanisms that might underlie diversity-dependent dynamics at
the clade level remain poorly known and controversial (Rabosky
2010b; Wiens 2011), but substantial evidence for such dynamics
comes from analyses of diversification patterns in species-level
phylogenies (Phillimore and Price 2008; Rabosky and Lovette
2008a; Rabosky and Glor 2010; Etienne et al. 2011). If diversity
dependence underlies the decoupling of age and species richness
in salamanders, then we should observe a general slowdown in
the rate of speciation through time within salamander clades.
We used the time-calibrated plethodontid phylogeny of
191 species from Adams et al. (2009) to test whether rates
of species diversification have slowed through time within the
15 focal clades, as predicted if such slowdowns can explain the
observed decoupling of age and species richness. Taxon sampling
within the focal clades is incomplete, but a majority of clades
(10/15) contain at least 50% of the known species (minimum:
32%; median: 58%). The analyses described below assume that
taxon sampling within clades is effectively random with respect
to phylogeny and that incompleteness reflects a lack of DNA samples for missing species rather than a targeted effort to include the
most phylogenetically divergent subset of lineages.
To test for overall departures from a constant speciation
model of clade diversification, we computed the γ-statistic (Pybus
and Harvey 2000) for each clade. Negative values of this statistic imply an excess of early speciation events in the phylogeny
relative to the expected number under a pure birth (constant rate)
model of diversification. Due to incomplete taxon sampling, we
generated null distributions of γ for each clade through simulation (Monte Carlo constant rate [MCCR] test; Pybus and Harvey
2000).
We then fitted time-varying models of speciation to each salamander clade using maximum likelihood (Rabosky and Lovette
2008b; Rabosky and Glor 2010; Morlon et al. 2011). Specifically,
we modeled speciation rates within clades as
λ(t) = λ0 ekt ,
where λ0 is the initial speciation rate, k is a rate change
parameter, and t is elapsed time from the initial (crown) speciation
event in the clade. This is a particularly useful model in the present
context, as it can approximate a linear diversity-dependent change
1812
EVOLUTION JUNE 2012
in the speciation rate through time (Quental and Marshall 2009).
To visualize the pattern of lineage accumulation under the fitted
models, we computed the expected number of lineages as
t
N (t) = 2 exp
λ(t)dt .
0
We accounted for incomplete sampling using the sampling
model developed by Morlon et al. (2011) for the time-varying
birth–death process. This approach enabled us to compute the likelihood of our phylogenetic data under any time-varying model of
speciation, conditional on our assumed level of species sampling.
We compared these fitted models to a simple constant rate speciation process in an Akaike information criterion (AIC) framework.
We found that including extinction in the model did not appreciably improve our ability to model the data; moreover, estimated
extinction rates tended toward zero for most clades. Given these
results and general problems associated with the estimation of extinction from molecular phylogenies (Rabosky 2009c, 2010a), we
present only results for the two-parameter time-varying speciation
model and one-parameter CR speciation model.
CORRELATES OF PHENOTYPIC EVOLUTIONARY
RATES
We used PGLS to assess the predictive ability of (1) CR diversification rate estimates and (2) log-transformed species richness in explaining the variation in phenotypic evolutionary rates
across salamander clades. If the dynamics of species richness
within clades is characterized by strong diversity dependence,
then species richness itself emerges as an appropriate response
variable in phylogenetic comparative analyses as an estimate
of the total time-integrated diversification experienced by clades
(Rabosky 2009b, 2010b). Alternatively, one can view the standing
richness of clades as an estimate of their carrying capacity. However, there is no need to ascribe a particular causal mechanism to
this relationship: we are ultimately interested in the factors that
determine clade richness, and if clade richness is not a function of
time, there is little justification for “correcting for time” by transforming species richness into a CR diversification rate estimate.
Our PGLS models included a parameter for Pagel’s lambda (denoted here by ), a scalar multiplier of the off-diagonal elements
of the phylogenetic variance–covariance matrix that reflects the
amount of phylogenetic signal in the residuals of the relationship
between dependent and independent variables (Pagel 1997). We
compared the relative importance of the two predictor variables
(log-transformed richness and net diversification rate) for rates
of both size and shape evolution using the Akaike information
criterion. All analyses and simulations were conducted in the R
statistical/programming environment. Source code and data underlying the analyses presented here have been deposited in the
Dryad online data repository (doi:10.5061/dryad.vt41c78j).
S P E C I E S D I V E R S I F I C AT I O N A N D P H E N OT Y P I C E VO L U T I O N
A
Corr (age*diversity)
1.0
0.5
0.0
-0.5
0.0
B
0.2
0.4
0.6
0.8
1.0
0.2
0.4
0.6
0.8
1.0
Corr (age*diversity)
1.0
0.5
0.0
-0.5
0.0
Relative extinction rate
Expected rank-order correlation between clade age and
species richness for plethodontid salamanders as a function of dif-
Figure 3.
ferent relative extinction rates for (A) phylogenetic and (B) uncorrelated lognormal models of rate variation among clades. Arrow
denotes observed age-diversity correlation for the salamander
dataset. Both phylogenetic and nonphylogenetic models predict
substantial positive relationships between clade age and species
richness.
Results
We found that among-clade variation in net diversification rates is
unlikely to explain the lack of relationship between clade age and
species richness in plethodontid salamanders. Posterior predictive simulations under the uncorrelated and phylogenetic models
of diversification rate variation indicate that strong positive correlations between age and species richness are expected for these
data under all scenarios we considered that allowed rates to vary
among clades (Fig. 3). For the phylogenetic model of rate variation
(Fig. 3A), the probability of observing an age-diversity correlation
equal to or less than that observed for salamanders (Spearman’s
ρ = –0.15) is less than 0.025 under all values of ε ≤ 0.8, and is
only somewhat more likely (P = 0.076) under the extreme case
where speciation and extinction are approximately balanced (ε =
0.99). Under the uncorrelated lognormal model (Fig. 3B), the observed age-diversity correlation is even less likely (P < 0.005
for ε ≤ 0.8; P = 0.039 for ε = 0.99). Our results imply that CR
estimators of net diversification provide an inaccurate measure of
clade diversification history for this group.
When we considered within-clade patterns of speciation, we
found a striking signature of declining diversification rates across
most plethodontid clades (Table 1). Observed γ-statistics were
consistently less than zero, with most clades (11/15) showing
at least marginally significant (P < 0.10) departures from a CR
speciation process after accounting for incomplete taxon sampling. Similar results were obtained for model-based analyses of
speciation dynamics, where 13/15 clades were characterized by
AIC > 0 in favor of a time-varying speciation model. For each
clade, the fitted time-varying models predict temporally declining
rates of speciation and lineage accumulation (Fig. 4). Total AIC
evidence across all 15 clades strongly favors a model with cladespecific and temporally declining rates of speciation, relative to
a model with clade-specific but time-invariant rates of speciation
(cumulative AIC = 29.2).
We found a significant positive relationship between logtransformed species richness and both phenotypic rate measures
(Table 2; Fig. 5). However, we found no relationship between
CR estimates of net diversification rate and rates of size or shape
evolution, consistent with previous results (Adams et al. 2009).
For rates of size and shape evolution, AIC scores favor logtransformed species richness over CR diversification estimates
as a predictor of phenotypic rates (size rate: AIC = 5.2; shape
rate: AIC = 5.1). CR estimates of net diversification rate given
in Table 2 assume a relative extinction rate (μ/λ) of 0.45 (after
Adams et al. 2009), but similar evidence favoring models with logtransformed richness was obtained when the relative extinction
rate was treated as a free parameter to be estimated during PGLS
analyses (size rate: AIC = 5.3; shape rate: AIC = 4.3). The
relative extinction rate estimated from this analysis approached
1.0, but this estimate is likely to reflect model misspecification.
Rabosky (2010b) documented severe statistical pathologies associated with the estimation of relative extinction rates from age
and richness data when the assumptions of a time-homogeneous
birth–death process are violated.
Discussion
These results indicate that log-transformed species richness is correlated with rates of phenotypic evolution in plethodontid salamanders and provide one of the first direct tests of the hypothesis
that species diversification and morphological evolutionary rates
can be coupled. By explicitly testing the validity of the CR diversification model, we found that CR estimators do not provide
a valid summary of clade diversification histories in salamanders. As species richness in clades is not a function of time-fordiversification, it is not appropriate to “correct” for clade age by
computing CR estimates of net diversification rates for each clade.
Consistent with this possibility, we found that species richness
itself is a better predictor of phenotypic evolutionary rates than
CR estimates of net diversification (Table 2).
EVOLUTION JUNE 2012
1813
DA N I E L L . R A B O S K Y A N D D E A N C . A DA M S
Table 1. Diversification through time in plethodontid salamander clades. Clade numbers correspond to those in Adams et al. (2009).
MCCR P-value is the one-tailed probability of the observed γ-statistic under the null hypothesis of constant speciation through time.
AICCR and AICVR are AICs for the constant-rate (CR) and variable-rate models of speciation, respectively.
Clade
(1) Desmognathus and Phaeognathus
(2) Aneides
(3) Western Plethodon
(4) P. cinereus group
(5) P. wehrlei-welleri group
(6) P. glutinosus group
(7) Gyrinophilus, Pseudotriton, and
Stereochilus
(8) Eurycea
(9) Nototriton
(10) Oedipina
(11) Chiropterotriton
(12) Pseudoeurycea clade
(13) Bolitoglossa, subgenus Eladinea
(14) Bolitoglossa, subgenera Magnadigita,
Oaxakia, and Pachymandra
(15) Bolitoglossa, subgenera Bolitoglossa,
Mayamandra, and Nanotriton
Species in clade/
species in tree
γ
MCCR
P-value
AICCR
AICVR
AIC
37/12
6/5
9/6
10/7
7/6
28/18
7/4
–1.9
–1.37
–1.4
–1.53
–0.99
–2.48
–2.17
0.16
0.03
0.07
0.05
0.08
0.02
<0.01
87.08
36.47
44.54
44.4
39.46
109.95
27.47
88.64
34.69
42.14
43.75
38.96
105.74
17.72
–1.56
1.78
2.4
0.65
0.5
4.21
9.75
36/17
13/5
25/10
12/7
51/32
46/15
25/19
–0.92
–0.37
–1.74
–2.47
–2.03
–2.69
–2.65
0.42
0.45
0.11
<0.01
0.07
0.04
<0.01
118.87
28.59
63.42
45.19
215.13
88.47
121.97
120.08
29.97
62.09
38.49
214.47
88.15
117.86
–1.21
–1.38
1.33
6.7
0.66
0.32
4.11
17/10
–1.83
0.06
64.01
63.03
0.98
The simple Brownian motion estimates of phenotypic evolutionary rates we consider here potentially suffer from the same
limitations as CR estimators of species diversification rates. For
example, constraints on phenotypic divergence can lead to negative correlations between phenotypic evolutionary rates and clade
age. If a set of clades occupy similar volumes of morphological
space but vary substantially in age, then we might estimate low
rates of evolution for old clades and fast rates for young clades.
Negative correlations between phenotypic evolutionary rates and
clade age (or timescale of measurement) have been reported in
several previous studies (Gingerich 2001; Ackerly 2009), suggesting the possibility that the phenotypic rates we consider here
might be confounded with clade age. However, our analyses of the
relationship between clade age, phenotypic rates, and phenotypic
disparity suggest that these rates are not confounded by clade age
(Fig. 1; see Materials and Methods).
It is possible that error in the estimation of clade age
could weaken or even eliminate a “true” positive correlation
between age and diversity. We suggest that this is unlikely
for several reasons. First, we find that phenotypic disparity increases with clade age, as expected under a model of trait evolution by Brownian motion (Fig. 1C). There is no reason to
expect this result if estimated clade ages are only weakly correlated with true clade ages. Second, through our analysis of
species-level diversification patterns, we have provided independent evidence that speciation rates have declined through
1814
EVOLUTION JUNE 2012
time in most clades. Such temporal decelerations in diversification rates can lead to a decoupling between age and diversity
at the clade level (Rabosky 2009b). In a more general sense,
we note that fossil-based clade ages are sometimes uncorrelated
with species richness (Magallon and Sanderson 2001; Rabosky
2009b), further supporting the notion that this is a real biological
phenomenon.
A number of causal mechanisms may underlie the patterns
we have documented in this study. One interpretation of these
results is that clades with higher rates of phenotypic evolution
diversify more than low-rate clades. If there are strong diversitydependent controls on species richness within clades, then clades
with higher phenotypic rates may occupy increasingly broad regions of ecological space. If the number of species is determined
by the size of the “adaptive zone” (Simpson 1953) occupied by
the clade, and if the overall size of the adaptive zone is itself a
function of the evolvability or versatility of the clade (Vermeij
1973a; Liem and Osse 1975), then species-rich clades should
be characterized by higher rates of ecological innovation, which
may in turn be approximated by the phenotypic diversification
rates we measured in salamanders. This assumes that the rates
of phenotypic evolution we have measured are associated with
ecological diversification, and additional data are needed to test
this hypothesis.
However, we cannot yet determine whether phenotypic evolutionary rates may be causally related to species richness. One
S P E C I E S D I V E R S I F I C AT I O N A N D P H E N OT Y P I C E VO L U T I O N
4
(1)
(2)
(3)
(4)
(5)
0.6
0.3
0
4
0.0
(6)
(7)
(8)
(9)
(10)
0.6
2
0.3
0
4
0.0
(11)
(12)
(13)
(14)
Speciation rate
Log number of species (estimated)
2
(15)
0.6
2
0.3
0
0.0
0.0
0.5
1.0
0.0
0.5
1.0
0.0
0.5
1.0
0.0
0.5
1.0
0.0
0.5
1.0
Relative divergence time
Figure 4. Maximum likelihood estimates of the rate of speciation through time (black lines) for 15 plethodontid salamander clades.
Gray lines denote estimated log-transformed diversity through time under the fitted model. Plot numbers correspond to clade indices in
Table 1.
Statistical summary of PGLS regressions between two measures of species diversification (net diversification rate or logtransformed species richness) and morphological evolutionary rates. i is the AIC weight for the model, and (“lambda”) is the
Table 2.
maximum likelihood estimate of Pagel’s (1997) parameter for phylogenetic signal.
Model
AIC
i
β
t
P
Size rate by net diversification rate
Size rate by species richness
Shape rate by net diversification rate
Shape rate by species richness
–66.1
–71.3
–173.2
–178.3
0.07
0.93
0.07
0.93
0
0.36
1
1
0.105
0.017
0.0044
0.0005
0.73
2.057
1.13
2.979
0.479
0.031
0.279
0.02
Statistically significant results are in bold.
alternative interpretation of these results is that species richness
itself drives higher rates of phenotypic evolution. This might
be expected if ecological character displacement is more common in species-rich clades, perhaps due to denser packing in
geographic and ecological space. Competitive interactions are
well known to have shaped patterns of community composition
(Adams 2007) and morphological variation in plethodontid communities (Adams and Rohlf 2000; Adams 2004, 2010). However,
causality in this scenario is also unclear: it is also possible that
lineages with greater capacity for ecological character displacement will contain more species, particularly if species persistence
and/or speciation itself is facilitated by ecological divergence from
other species. This is especially relevant when considering the effects of range expansions on speciation and species persistence.
Geographic range expansions may be a critical part of “successful” speciation, but they may require that recently separated al-
lospecies undergo ecological or morphological divergence before
sympatry is possible (Rundell and Price 2009).
The lack of correlation between clade age and species richness suggests the intriguing possibility of diversity-dependent
control of clade size in salamanders, consistent with previous evidence for declining speciation rates observed in several plethodontid subclades (Kozak et al. 2006; Kozak and Wiens 2010). Our
finding of parallel declines in the rate of speciation through time
within plethodontid clades provides further support for this hypothesis (Table 1; Fig. 4). Previous studies have speculated that
diversity-dependent declines in net speciation rates might underlie
the decoupling of age and species richness reported in other groups
(Ricklefs and Renner 1994; Ricklefs 2007; Rabosky 2009a). To
our knowledge, this is the first study to demonstrate correspondence of within-clade patterns of speciation through time and
age-diversity relationships at the clade level.
EVOLUTION JUNE 2012
1815
DA N I E L L . R A B O S K Y A N D D E A N C . A DA M S
A
nisms that underlie the age-diversity relationship in salamanders
and other taxa will require greater integration of phylogenetic,
ecological, geographic, and paleontological data.
0.10
Size rate
0.08
0.06
A CAUTION FOR MODEL-BASED INFERENCE
0.04
0.02
0.00
2
3
4
Log (richness)
B
Shape rate
0.002
0.001
0.000
2
3
4
Log (richness)
Relationship between rates of morphological evolution and total species diversification (log-transformed richness) for
15 clades of plethodontid salamanders. (A) Relationship between
Figure 5.
rate of body size evolution and species richness. (B) Relationship
between rate of shape evolution and species richness. Both phenotypic evolutionary rates are significantly and positively associated
with species richness (size rate: P = 0.033; shape rate: P = 0.019).
We are only beginning to understand the population ecological mechanisms that might define a macroevolutionary “adaptive
zone” or “carrying capacity” at the clade level. Such a carrying capacity is clearly more than the simple sum of local-scale
niches that are occupied by species within a clade and potentially includes geographic opportunities for speciation (Kisel and
Barraclough 2010), the effects of species interactions on range
size and range expansions (Price and Kirkpatrick 2009), and feedbacks between local and regional diversity. A lack of relationship
between clade age and species richness does not constitute a strong
test for diversity dependence. However, given that we have also
found a signal of declining diversification rates within salamander clades, diversity dependence remains an important candidate
process to explain this pattern. If the decoupling between age and
species richness is merely an artifact of the way higher taxa are delimited (Rabosky 2010b), then we would not expect to observe (1)
correlations between phenotypic covariates of putative ecological
significance (shape and size rates) and species diversification, and
(2) a consistent signal of declining speciation through time within
clades (Fig. 4). Regardless, more effective tests of the mecha-
1816
EVOLUTION JUNE 2012
Using an AIC framework, we found that log-transformed species
richness is a better predictor of both size and shape evolutionary rates than CR estimates of net diversification rate (Table 2).
However, although AIC comparisons provide corroborative evidence favoring one model over another, in some instances model
selection can be positively misleading; for example, if the independent variable (diversification rate or richness) and dependent
variable (ecological or phenotypic covariate) are related by virtue
of secondary correlations with some unknown factor. This is especially likely in the analyses of evolutionary rates, where time can
potentially confound both the predictor and response variables.
Consider a simple scenario where clade richness is independent of
clade age, perhaps due to diversity-dependent regulation of speciation and extinction. CR estimates of net diversification for such
clades may be devoid of biological meaning but will nonetheless show negative correlations with clade age (Rabosky 2009a);
they will thus be correlated with any time-dependent ecological
or phenotypic covariates.
This is a serious concern, partly because many potential correlates of species diversification show real or artifactual correlations with the timescale over which they are measured (including phenotypic evolutionary rates [Kurten 1959; Gingerich 2001;
Ackerly 2009] and molecular evolutionary rates [Ho et al. 2011]).
These problems are exacerbated if researchers interpret correlations between “net diversification rates” and ecological covariates
as evidence that those rates are causal with respect to species richness. Whether such rates can potentially explain species richness
must be addressed independently, as we have done here for salamanders.
Summary
The results presented here indicate that assumptions about models of diversity regulation can have profound consequences for
interpreting macroevolutionary patterns. Adams et al. (2009) estimated salamander diversification rates under the assumption
that diversity has increased continuously through time, finding no
correlation between diversification rates and rates of phenotypic
evolution. By applying new tools for modeling diversification dynamics within clades, we found that species richness is itself a
more appropriate variable in “downstream” phylogenetic comparative analyses. Species richness across 15 clades of plethodontid
salamanders is correlated with rates of body size and shape evolution, consistent with the hypothesis that phenotypic evolutionary
rates promote species diversification. Although we cannot yet
S P E C I E S D I V E R S I F I C AT I O N A N D P H E N OT Y P I C E VO L U T I O N
establish the direction of causality, our results are consistent with
a number of models that postulate coupling between patterns of
phenotypic evolution and species diversification.
ACKNOWLEDGMENTS
We thank J. Huelsenbeck, L. Mahler, A. Rabosky, and R. Ricklefs for
comments and/or discussion that improved the manuscript; and H. Morlon
for sharing R code. This research was supported in part by the Miller
Institute for Basic Research in Science at the University of California,
Berkeley and by NSF DEB-1118884 (to DCA).
LITERATURE CITED
Ackerly, D. 2009. Conservatism and diversification of plant functional traits:
evolutionary rates versus phylogenetic signal. Proc. Nat. Acad. Sci.
U.S.A. 106:19699–19706.
Adams, D. C. 2004. Character displacement via aggressive interference in
appalachian salamanders. Ecology 85:2664–2670.
———. 2007. Organization of Plethodon salamander communities: guildbased community assembly. Ecology 88:1292–1299
———. 2010. Parallel evolution of character displacement driven by competitive selection in terrestrial salamanders. BMC Evol. Biol. 10:1–10.
Adams, D. C., and F. J. Rohlf. 2000. Ecological character displacement in
Plethodon: biomechanical differences found from a geometric morphometric study. Proc. Nat. Acad. Sci. U.S.A 97:4106–4111.
Adams, D. C., C. M. Berns, K. H. Kozak, and J. J. Wiens. 2009. Are rates of
species diversification correlated with rates of morphological evolution?
Proc. R. Soc. Lond. B 276:2729–2738.
Alfaro, M. E., F. Santini, C. Brock, H. Alamillo, A. Dornburg, D. L. Rabosk,
G. Carnevale, and L. J. Harmon. 2009. Nine exceptional radiations plus
high turnover explain species diversity in jawed vertebrates. Proc. Nat.
Acad. Sci. U.S.A. 106:13410–13414.
Alroy, J. 2008. The dynamics of origination and extinction in the marine fossil
record. Proc. Nat. Acad. Sci. U.S.A. 105:11536–11542.
Atchley, W. R., C. T. Gaskins, and D. Anderson. 1976. Statistical properties
of ratios .1. Empirical results. Syst. Zool. 25:137–148.
Bailey, N. T. J. 1964. The elements of stochastic processes with applications
to the natural sciences. Wiley, New York.
Bokma, F. 2003. Testing for equal rates of cladogenesis in diverse taxa.
Evolution 57:2469–2474.
Drummond, A. J., S. W. Y. Ho, M. J. Phillips, and A. Rambaut. 2006.
Relaxed phylogenetics and dating with confidence. PLoS Biol. 4:
699–710.
Eastman, J. M., M. E. Alfaro, P. Joyce, A. L. Hipp, and L. J. Harmon. 2011. A
novel comparative method for identifying shifts in the rate of character
evolution on trees. Evolution 65:3578–3589.
Eldredge, N., and S. J. Gould. 1972. Punctuated equilibria: an alternative to
phyletic gradualism. Pp. 182–215 in T. J. M. Schopf, ed. Models in
paleobiology. Freeman Cooper, San Francisco.
Etienne, R. S., B. Haegeman, T. Stadler, T. Aze, P. N. Pearson, A. Purvis, and
A. B. Phillimore. 2011. Diversity-dependence brings molecular phylogenies closer to agreement with the fossil record. Proc. R. Soc. Lond. B
doi: 10.1098/rspb.2011.1439.
Gingerich, P. D. 2001. Rates of evolution on the time scale of the evolutionary
process. Genetica 112–113:127–144.
Harmon, L. J., J. A. Schulte, A. Larson, and J. B. Losos. 2003. Tempo and
mode of evolutionary radiation in iguanian lizards. Science 301:961–
964.
Heard, S. B., and D. L. Hauser. 1995. Key evolutionary innovations and their
ecological mechanisms. Hist. Biol. 10:151–173.
Ho, S. W. Y., R. lanfear, L. Bromham, M. J. Phillips, J. Soubrier, A. G. Rodrigo,
and A. Cooper. 2011. Time-dependent rates of molecular evolution. Mol.
Ecol. 20:3087–3101.
Kisel, Y., and T. G. Barraclough. 2010. Speciation has a spatial scale that
depends on levels of gene flow. Am. Nat. 175:316–334.
Kozak, K. H., and J. J. Wiens. 2010. Accelerated rates of climatic-niche
evolution underlie rapid species diversification. Ecol. Lett. 13:1378–
1389.
Kozak, K. H., D. W. Weisrock, and A. Larson. 2006. Rapid lineage accumulation in a non-adaptive radiation: phylogenetic analysis of diversification
rates in eastern North American woodland salamanders (Plethodontidae:
Plethodon). Proc. R. Soc. Lond. B 273:539–546.
Kozak, K. H., R. W. Mendyk, and J. J. Wiens. 2009. Can parallel diversification occur in sympatry? Repeated patterns of body-size evolution in
coexisting clades of North American salamanders. Evolution 63:1769–
1784.
Kurten, B. 1959. Rates of evolution in fossil mammals. Cold Spring Harb.
Symp. Quant. Biol. 24:205–215.
Liem, K. F., and J. W. M. Osse. 1975. Biological versatility, evolution, and
food resource exploitation in African cichlid fishes. Am. Zool. 15:
427–454.
Magallon, S., and M. J. Sanderson. 2001. Absolute diversification rates in
angiosperm clades. Evolution 55:1762–1780.
Mahler, D. L., L. J. Revell, R. E. Glor, and J. B. Losos. 2010. Ecological
opportunity and the rate of morphological evolution in the diversification
of greater Antillean anoles. Evolution 64:2731–2745.
Martin, C. H., and P. C. Wainwright. 2011. Trophic novelty is linked to exceptional rates of morphological diversification in two adaptive radiations
of Cyprinodon pupfish. Evolution 65:2197–2212.
Martins, E. P., and T. A. Hansen. 1997. Phylogenies and the comparative
method: a general approach to incorporating phylogenetic information
into the analysis of interspecific data. Am. Nat. 149:646–667.
McInnes, L. M., C. D. L. Orme, and A. Purvis. 2011. Detecting shifts in
diversity limits from molecular phylogenies: what can we know? Proc.
R. Soc. Lond. B 278:3294–3302.
McPeek, M. A. 2008. Ecological dynamics of clade diversification and community assembly. Am. Nat. 172:E270–E284.
McPeek, M. A., L. Shen, J. Z. Torrey, and H. Farid. 2008. The tempo and
mode of 3-dimensional morphological evolution in male reproductive
structures. Am. Nat. 171:E158–E178.
Morlon, H., T. L. Parsons, and J. B. Plotkin. 2011. Reconciling molecular phylogenies with the fossil record. Proc. Natl. Acad. Sci. U.S.A.
108:16327–16332.
Pagel, M. 1997. Inferring evolutionary processes from phylogenies. Zool.
Scripta 26:331–348.
Parent, C. E., and B. J. Crespi. 2009. Ecological opportunity in adaptive radiation of Galapagos endemic land snails. Am. Nat. 174:898–
905.
Phillimore, A. B., and T. D. Price. 2008. Density dependent cladogenesis in
birds. PLoS Biol. 6:e71.
Price, T. D., and M. Kirkpatrick. 2009. Evolutionarily stable range limits set
by interspecific competition. Proc. R. Soc. Lond. B 276:1429–1434.
Purvis, A. 2004. Evolution – How do characters evolve? Nature 432:1.
Pybus, O. G., and P. H. Harvey. 2000. Testing macro-evolutionary models using incomplete molecular phylogenies. Proc. R. Soc. Lond. B 267:2267–
2272.
Pyron, R. A., and F. T. Burbrink. 2011. Extinction, ecological opportunity,
and the origins of global snake diversity. Evolution 66:163–178.
Quental, T. B., and C. R. Marshall. 2009. Extinction during evolutionary
radiations: reconciling the fossil record with molecular phylogenies.
Evolution 63:3158–3167.
EVOLUTION JUNE 2012
1817
DA N I E L L . R A B O S K Y A N D D E A N C . A DA M S
Rabosky, D. L. 2009a. Ecological limits and diversification rate: alternative
paradigms to explain the variation in species richness among clades and
regions. Ecol. Lett. 12:735–743.
———. 2009b. Ecological limits on clade diversification in higher taxa. Am.
Nat. 173:662–674.
———. 2009c. Heritability of extinction rates links diversification patterns in
molecular phylogenies and fossils. Syst. Biol. 58:629–640.
———. 2010a. Extinction rates should not be estimated from molecular phylogenies. Evolution 64:1816–1824.
———. 2010b. Primary controls on species richness in higher taxa. Syst.
Biol. 59:634–645.
Rabosky, D. L., and R. E. Glor. 2010. Equilibrium speciation dynamics in a
model adaptive radiation of island lizards. Proc. Natl. Acad. Sci. U.S.A.
107:22178–22183.
Rabosky, D. L., and I. J. Lovette. 2008a. Density-dependent diversification in
North American wood warblers. Proc. R. Soc. Lond. B 275:2363–2371.
———. 2008b. Explosive evolutionary radiations: decreasing speciation or
increasing extinction through time? Evolution 62:1866–1875.
Raup, D. M. 1985. Mathematical models of cladogenesis. Paleobiology 11:42–
52.
Raup, D. M., S. J. Gould, T. J. M. Schopf, and D. Simberloff. 1973. Stochastic
models of phylogeny and evolution of diversity. J. Geol. 81:525–542.
Ricklefs, R. E. 2004. Cladogenesis and morphological diversification in
passerine birds. Nature 430:338–341.
———. 2006. Time, species, and the generation of trait variance in clades.
Syst. Biol. 55:151–159.
———. 2007. Estimating diversification rates from phylogenetic information.
Trends Ecol. Evol. 22:601–610.
Ricklefs, R. E., and S. S. Renner. 1994. Species richness within families of
flowering plants. Evolution 48:1619–1636.
Rosenzweig, M. L. 1975. On continental steady states of species diversity.
Pp. 121–140 in M. L. Cody and J. M. Diamond, eds. Ecology and
evolution of communities. Belknap, Cambridge, U.K.
1818
EVOLUTION JUNE 2012
Rundell, R. J., and T. D. Price. 2009. Adaptive radiation, nonadaptive radiation,
ecological speciation and nonecological speciation. Trends Ecol. Evol.
124:394–399.
Sepkoski, J. J. 1978. A kinetic model of Phanerozoic taxonomic diversity I.
Analysis of marine orders. Paleobiology 4:223–251.
Simpson, G. G. 1953. The major features of evolution. Columbia Univ. Press,
New York.
Slater, G. J., S. A. Price, F. Santini, and M. E. Alfaro. 2010. Diversity versus
disparity and the radiation of modern cetaceans. Proc. R. Soc. Lond. B
277:3097–3104.
Thorne, J. L., H. Kishino, and I. S. Painter. 1998. Estimating the rate of
evolution of the rate of molecular evolution. Mol. Biol. Evol. 15:1647–
1657.
Vamosi, J. C., and S. M. Vamosi. 2010. Key innovations within a geographical
context in flowering plants: towards resolving Darwin’s abominable
mystery. Ecol. Lett. 13:1270–1279.
Venditti, C., A. Meade, and M. Pagel. 2011. Multiple routes to mammalian
diversity. Nature 479:393–396.
Vermeij, G. J. 1973a. Adaptation, versatility, and evolution. Syst. Zool.
22:466–477.
———. 1973b. Biological versatility and earth history. Proc. Natl. Acad. Sci.
U.S.A. 70:1936–1938.
Walker, T. D., and J. W. Valentine. 1984. Equilibrium models of evolutionary
species diversity and the number of empty niches. Am. Nat. 124:887–
899.
Weir, J. T. 2006. Divergent timing and patterns of species accumulation in
lowland and highland neotropical birds. Evolution 60:842–855.
Wiens, J. J. 2007. Global patterns of diversification and species richness in
amphibians Am. Nat. 170:S86–S106.
———. 2011. The causes of species richness patterns across space, time, and
clades and the role of “ecological limits”. Q. Rev. Biol. 86:75–96.
Associate Editor: J. Vamosi